Sreeyuth Chowdary — Software Developer

About

I'm a Software Developer at Northwest Missouri State University's Learning and Teaching Center and a Computer Science student (GPA 3.6/4.0, Dec 2026). I build production-grade systems that combine AI, backend engineering, and user-focused product design.

Recently, I built NWMSU Agent Core, an AI-powered campus assistant that integrates multiple LLMs, tool-augmented workflows, and secure university data access across web and Microsoft Teams. My core strength is shipping reliable end-to-end systems, from orchestration and APIs to deployment and observability.

I'm currently seeking Summer 2026 software engineering internships where I can contribute to high-impact product teams building intelligent, scalable software.

Association for Computing Machinery

Conference Presentation

CCSC Central Plains 2025 — Presented research poster on Enhancing Adjunct Faculty Performance Prediction by Applying SMOTE Techniques at Drake University, Des Moines, IA.

Work Experience

Software Developer

Learning and Teaching Center, Maryville, MO

Sep 2024 - Present

Built NWMSU Agent Core, an AI-powered campus assistant using FastAPI, React, TypeScript, Docker, Redis, PostgreSQL, OpenAI, Gemini, Microsoft Graph, and Canvas API to centralize student-facing academic support workflows.
Designed a multi-LLM, tool-augmented architecture with MCP-based microservices for Canvas, Microsoft Graph, document processing, and university web search.
Engineered an async orchestration layer for tool routing, authentication, conversation state management, and real-time response streaming across web and Microsoft Teams.
Built secure OAuth-based integrations and session-managed access controls for protected Canvas and Microsoft data workflows.
Managed Docker-based multi-service deployments with Prometheus, Grafana, and Loki for observability, monitoring, logging, and debugging.
Developed an Adjunct Faculty Management System for role-based administrative workflows and continued maintaining internal university applications through feature delivery, bug fixes, reliability improvements, and CI/CD automation.

Projects

Podcast Brain

Architected and shipped a full-stack AI podcast platform that combines transcription, semantic retrieval, and streaming Q&A. Built a production RAG pipeline with transcript chunking, OpenAI embeddings, and PostgreSQL pgvector search, delivering ~120ms top-5 retrieval latency and ~1.8s time-to-first-streamed response.

Next.js 14 TypeScript Supabase OpenAI pgvector

Live GitHub

Gesture Code

Developed and published a VS Code extension for hands-free editor control using webcam-based gesture recognition. Implemented MediaPipe Hands classification with 21 landmarks, handedness-aware detection, and stability logic, then engineered a Node.js bridge server to work around VS Code CSP limits for WebAssembly-powered inference.

TypeScript MediaPipe Node.js VS Code API

Open VSX GitHub

Pose Fitness Coach

Built a real-time computer vision fitness coach that tracks 33 body landmarks and counts reps across 8 exercises with robust state-machine transitions. Designed a modular OOP tracking framework and validated movement logic with a 33-test pytest suite to improve reliability and correctness.

Python MediaPipe OpenCV pytest

Live GitHub